Monoclonal antibodies selectively bind tumor cell differentiating antigens in vitro and in vivo. Natural effector mechanisms often do not mediate killing of monoclonal antibody bound cells so we have devised methods of linking extremely toxic proteins to the antibodies to selectively kill tumor cells. Two methods of coupling toxic proteins, like ricin to antibodies, have been used to kill antigen positive cells in vitro. Ricin has two subunits, the A subunit blocks protein synthesis when in the cytosol and the B subunit binds galactose groups on all cell surfaces but also facilitates the transport of ricin A chain to the cytosol. 1) Linkage of the ricin A chain to antibodies yields reagents with low non-target toxicity but target cell toxicity too slow for in vivo applications; 2) Linkage of intact ricin to antibodies results in very potent target cell toxicity but the non- target cell killing must be prevented by a ligand which blocks ricin B chain binding to cells. This has limited its application to in vitro situations where 100 mM lactose can block ricin binding. We have succeeded in developing several new approaches to apply immunotoxins in vivo. 1) Cloning of toxins then altering their structure at the gene level to decrease non-target cell toxicity; 2) Chemical modification of ricin to determine the location of the ricin galactose binding site and to possibly improve efficacy of ricin linked to antibodies; 3) Develop new ways to block the non- target cell toxicity of ricin in vivo. We have discovered a monoclonal antibody which blocks the ricin galactose binding site similar to lactose; 4) Intrathecal administration of monoclonal antibodies allows a new way to image brain tumors; 5) Intrathecal administration of immunotoxins for therapy of brain tumors that kill 2 to 5-logs of tumor cells in animal models; and 6) Preparation for clinical trials of human brain tumor patients.